Meta is making a significant investment in Nvidia’s renowned computer chips, a crucial component in the field of artificial intelligence (AI) research and development. This financial commitment reflects the company’s dedication to advancing in AI technology, as indicated in a recent Instagram Reels post by Mark Zuckerberg, the CEO of Meta.
Zuckerberg’s announcement highlighted Meta’s ambitious plans for AI, emphasizing the need for an extensive compute infrastructure to support their future AI projects. By the close of 2024, this infrastructure is expected to incorporate an impressive 350,000 H100 graphics cards from Nvidia. The H100 is Nvidia’s latest offering in the high-performance computing market and has been in limited supply since its launch in late 2022.
While Zuckerberg did not disclose the current number of these graphics processing units (GPUs) acquired by Meta, their importance in AI research is undeniable. Raymond James analysts estimate the cost of each H100 unit to be between $25,000 and $30,000, though prices on platforms like eBay can surge above $40,000. If Meta were to purchase these GPUs at the lower end of this price spectrum, the expenditure would approach a staggering $9 billion.
This significant investment is a part of Meta’s broader strategy to enhance its compute capabilities. Zuckerberg mentioned that the company’s compute infrastructure would also encompass nearly 600,000 H100 equivalent units, factoring in other GPUs. In a remarkable step toward diversifying their AI hardware, Meta, alongside other tech giants like OpenAI and Microsoft, declared in December their intent to utilize AMD’s new Instinct MI300X AI computer chips.
The pursuit of artificial general intelligence (AGI) is a primary driver behind Meta’s need for these advanced computer chips. AGI, a futuristic form of AI that mirrors human-level intelligence, is a long-term vision for Meta, as stated by Zuckerberg. This field of research is not exclusive to Meta; other prominent organizations like OpenAI and Google’s DeepMind are also deeply involved in AGI research.
Yann LeCun, Meta’s chief scientist, underscored the critical role of GPUs in a media event held in San Francisco last month. He highlighted the direct correlation between the pursuit of AGI and the necessity for more GPUs. LeCun also commented on the competitive nature of the AI industry, likening it to a war, with Nvidia CEO Jensen Huang providing the essential “weapons” in the form of advanced GPUs.
Meta’s financial commitment to AI and computing expansion is further evidenced in their third-quarter earnings report. The report indicates that the company anticipates total expenses for 2024 to range between $94 billion and $99 billion, with AI and computing resources being the primary investment areas.
In addition to infrastructure investment, Zuckerberg shed light on Meta’s approach to AI development. He expressed the company’s intention to “open source responsibly” its forthcoming developments in “general intelligence.” This strategy aligns with Meta’s ongoing efforts with its Llama family of large language models. The company is currently training Llama 3 and is fostering closer collaboration between its Fundamental AI Research team (FAIR) and GenAI research team.
Following Zuckerberg’s announcement, LeCun, in a post on X, elaborated on the organizational changes within Meta’s AI divisions. He indicated that to expedite progress, FAIR is now operating alongside GenAI, the AI product division, as a sister organization. This restructuring is indicative of Meta’s commitment to integrating AI research with practical AI applications, thereby solidifying its position as a frontrunner in the AI domain.
In summary, Meta’s substantial investment in Nvidia’s GPUs, coupled with their strategic planning and organizational adjustments, underscores their deep commitment to advancing AI technology. As the company ventures into the realms of AGI and other AI applications, its focus on robust compute infrastructure, collaborative research, and open-source initiatives positions it as a key player in shaping the future landscape of artificial intelligence.